A customer books an appointment. Before that, she saw your Instagram ad, searched your name on Google a few days later, read two blog posts, and finally clicked a Google ad and filled in the form. Meta’s report claims the booking. Google Ads claims it. GA4 gives part of it to organic search. You paid for one booking, but three tools say they caused it.
That is the attribution problem, and every business that uses more than one channel runs into it. Marketing attribution explained simply: it is the set of rules a tool uses to decide which touchpoint gets credit for a result. This post covers the common models, why tools disagree even when they are working properly, and a practical way for a small business to judge channels without a data scientist.
Marketing attribution explained in practice
Most people do not see one ad and buy. They come across you several times, through different channels, before they act. Each of those moments is a touchpoint.
An attribution model is the rule for sharing credit among them. It might give everything to the last click, everything to the first, or split credit using some formula. Change the rule, and the same set of customers makes a different channel look best.
None of the models is the truth. Each is a way of looking at the same journeys. The useful question is not “which model is right?” but “which view helps me make this particular decision?”
The main models, from simple to automated
| Model | How it shares credit | Tends to favour |
|---|---|---|
| Last click | All credit to the final click before the result | Channels that close: brand searches, Google Ads, email |
| First click | All credit to the first known touch | Channels that introduce: social ads, blog content, display |
| Rule-based splits | Fixed shares, such as equal credit to each touch | Long journeys with several channels |
| Data-driven | Credit shared using patterns the platform finds in its own conversion data | Whatever patterns the platform detects, within what it can see |
GA4’s reports currently offer data-driven attribution and last-click options, and Google has removed several older rule-based models over time. Google Ads uses data-driven attribution for most new conversion actions. Meta uses its own attribution setting, which counts results within a set window after someone clicks an ad and, depending on the setting, after they view one.
Last click vs data-driven: the practical difference
Under last click, the final step always wins. If people usually find you on Instagram and then search your business name before booking, last click hands every booking to Google search and makes Instagram look useless. Cut Instagram on that evidence and, a few months later, fewer people may be searching your name.
Data-driven attribution tries to correct for that by noticing which touchpoints tend to appear in journeys that convert. It is more balanced in principle, but it has limits a small business should know:
- It works within one platform’s view. Google’s data-driven model sees Google’s touchpoints and GA4’s visits, not the Instagram ad someone scrolled past without clicking.
- It needs volume. With a modest number of conversions, the model has little to learn from.
- It is not transparent. You cannot see exactly why it gave a channel a particular share.
Last click is easy to understand and good for judging what closes sales. Data-driven gives more credit to earlier steps. Looking at both and noticing where they differ is often more useful than choosing one.
Why every tool claims the same customer
Even with identical models, tools disagree because they see different things:
- Each ad platform credits itself. Meta reports results after its own ads, Google Ads after its own ads. Neither subtracts the other.
- Views versus clicks. Meta can count a result after someone merely saw an ad. GA4 only sees visits to your site.
- Different time windows. A click weeks ago may count in one tool and have expired in another.
- Consent and blockers. Each tool loses a different set of visitors.
Other causes of mismatched numbers, such as clicks that never become visits, time zones and bots, are covered in our post on ad platform figures versus website analytics. Make sure tracking is not simply double counting before blaming attribution; our guide to conversions counted twice helps rule that out.
Where to see journeys in GA4
GA4’s advertising section includes a conversion paths view, showing common sequences of channels before a key event, and a comparison of how different models credit each channel. In the admin area, attribution settings control the model used in reports and how far back GA4 looks for touchpoints.
For most small businesses, an occasional look at conversion paths is enough. It shows whether enquiries usually come after a single visit or after several, and which channels tend to appear early. That alone changes how you judge channels like social media and content.
A practical approach for a small business
You do not need a perfect model. You need a sensible way to make budget decisions. This works for most small businesses:
- Use your own records as the total. The number of real enquiries and sales comes from your inbox, call log and order list, not from adding up platform reports.
- Ask how people found you. A short “how did you hear about us?” question on the form, or asked on the phone, captures what tools miss, including word of mouth and offline ads. For links you share yourself, labelling campaign links with UTM tags keeps those visits identifiable too.
- Give each channel a role. Search ads and brand searches close. Social ads and content introduce. Email and WhatsApp bring people back. Judge each against its role rather than against last click alone.
- Watch total results when you change something. If you pause a channel for a while and total enquiries hold steady, it was getting credit for work others were doing. If they fall, it mattered more than its report suggested.
- Compare trends, not single months. Small numbers swing. A quarter tells you more than a week.
This is the same thinking behind the monthly routine in our marketing analytics guide for small business owners. Attribution reports are useful context; your own records and simple tests make the decision.
Common attribution mistakes
- Adding Google Ads, Meta and GA4 conversions together and treating the sum as your total.
- Cutting awareness channels because last click gives them nothing.
- Trusting one platform’s view of its own value without checking against real results.
- Changing the attribution model halfway through a comparison, then comparing months that used different rules.
- Expecting a precise answer from a small number of conversions a month.
Frequently asked questions
Which attribution model should a small business use?
For day-to-day reports, the default in each tool is usually fine, which today is often data-driven. What matters more is not switching models midway through a comparison and checking platform claims against your own records. Looking at last click alongside it helps you see which channels close and which introduce.
Can I track customers across devices, like phone then laptop?
Only partly. Google and Meta can sometimes connect devices for signed-in users, but GA4 often treats a phone visit and a later laptop visit as two different users. That is one reason enquiry forms that ask how people found you remain useful alongside tracking.
Do offline ads like newspapers or flyers fit into attribution?
Not automatically, because tools only see digital touchpoints. A dedicated tracking phone number, a tagged QR code or a specific offer code for each offline ad brings some of that activity into view, and asking customers fills the rest.
Why did my Google Ads conversions change after an attribution update?
Changing the model, or Google moving an action to data-driven attribution, redistributes credit between campaigns and keywords, so individual figures shift even if total enquiries did not. Note the date of any change and compare totals from your own records before and after it.
Attribution is easier to live with when the tracking underneath it is clean and your own records are kept up to date. Our tracking and reporting setup focuses on that foundation rather than on complex models. For how each channel plays its part, see our overview of digital marketing for small businesses.
Next in this series: Google Tag Manager Preview Mode: Test Tags Safely
